gnrc_tcp: fix return on closed state #7069
Conversation
|
ACK & go |
Actually, there have been nightly builds for weeks. Just the results were boring enough I didn't bother sending the results somewhere. :) |
|
That mistake was mine 😢 however, even given more time (some hours ... a day or more) this might still have happened, especially if merged not by the same person. Both PRs were squashed, approved, and Murdock was all green. I haven't thought about triggering Murdock for the second PR again after merging the first one - though would have been better/correct to do here as both PRs touched same code - that obviously was a fault. Or what would have been the right procedure to avoid this for future reference? |
Why don't we do something like this: https://github.com/libgit2/libgit2#libgit2---the-git-linkable-library |
#6969 and #6997 were merged in rapid succession. However, #6969 introduced a new code path were
gnrc_tcp_close()would return early (with a value), while #6997 set the return type of that function tovoid. Because of this master is currently broken (I guess someone owes the community a box of beer ;-)). Murdock wasn't able to detect this because at build time neither of the PRs was merged into master. I guess it's high time for nightly builds, @kaspar030 ;-).